Trading and Arbitrage in Cryptocurrency Markets

39 Pages·2017·3.11 MB·English

Trading and Arbitrage in Cryptocurrency Markets
Igor Makarov Antoinette Schoar
LSE MIT Sloan
Gerzensee, October 10, 2018
Motivation
• Cryptocurrencieslikebitcoinarebuiltontheblockchaintechnology
• Allowsverificationofpaymentsintheabsenceofacentralizedcustodian
• Verificationisdonebydecentralized"miners"
• BitcoinwasoriginallyintroducedinapaperbyNakamoto(2008)and
cameintoexistencein2009
• FirstpaymenttransactiononMay22,2010: Abitcoinenthusiast(Laszlo
Hanyecz)bought2pizzasfor10,000bitcoins
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 2
Motivation (cont.)
• Sincethenthemarketforcryptocurrencieshasevolveddramatically
• Morethen1000“altcoins”tradedonmorethan100exchanges
• Marketcap: $500Batthepeak
• Academicresearchonbitcointradingisinitsbeginnings
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 3
This paper
• Asystematicanalysisofthetradingandefficiencyofcryptomarkets
• Cryptomarketsareidealsettingsforstudyingpricearbitrage:
• Marketsaresegmentedacrossmanycountriesandjurisdictions
• Many‘naive’investorsandfewlargesophisticatedinvestors(e.g.,DRW,
JumpTrading,orHehmeyerTrading)
• Blockchaintechnologyalleviatessomeconstraints(e.g.,capitalmobility)
butintroducesothers(thetransferofvaluebetweenexchangesissubject
toadelay)
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 4
Main results
• Historyofbitcoinexchangesmarkedbyrecurringepisodesof
arbitrageopportunitiesopeningandclosingagain
• ThetotalsizeofarbitrageprofitsfromDecember2017toFebruary2018
iswellabove$1billion
• Arbitrageopportunitiespersistforseveralhoursorevendaysandweeks
• Marketsegmentationmatters
• Arbitrageopportunitiesarelargeracrosscountries(orregions)than
withinthesamecountry
• Arbitragespreadsaremuchsmallerforexchangeratesbetweendifferent
cryptocurrenciescomparedtoexchangeratesbetweencryptocurrencies
andfiatcurrencies
• Arbitragespreadsacrosscountriesshowstrongco-movement
• Countrieswithhigheraveragebitcoinpremiumalsorespondmore
stronglytoperiodsof’buyingpressure’
• Pricedeviationsareasymmetric: BitcoinpriceinrestofworldisaboveUS
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 5
Main results (cont.)
• Bitcoinreturnsandarbitragespreadsvarywithnetorderflows
• Wedecomposesignedvolumeoneachexchangeintoacommon
componentandanidiosyncratic,exchange-specificcomponent
• Thecommoncomponentexplains80percentofthevariationinbitcoin
returns
• Buying10,000bitcoinsraisesreturnsby4%atthedailyfrequency
• Theidiosyncraticcomponentsoforderflowplayanimportantrolein
explainingthesizeofthearbitragespreadsbetweenexchanges
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 6
Overview
• Bitcoinprice: January1st,2016–February28,2018
14. Volume-weighted daily price (using trades on Bitstamp and Coinbase)

MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 7
1.3 Without China
Overview
• Dailybitcoinvolumetofiatcurrenciesin2017

Exchanges:
UMaSk:a CrovoiannbdaSsceh,o aBr,itTsratdaimngpa,n GdeArmbiitnraig, eKinraCkryepnto currencyMarkets 8
Japan: Bitflyer, Zaif, Quoine
Korea: Bithumb, Korbit
Hong Kong: Bitfinex
Europe: Kraken, Coinbase, Bitstamp
Tether: Poloniex, Binance, Bittrex
Data
• TickleveltradingdatafromKaiko,aprivatefirmthathasbeen
collectingtradinginformationaboutcryptocurrenciessince2014
• TheKaikodatacoverthe15largestandmostliquidexchanges:
Bitstamp,Kraken,BTCC,Bittrex,Coinbase,OkCoin,Bitfinex,Poloniex,
Bithumb,Gemini,Quoine,bitFlyer,Huobi,Binance,andZaif
• The15exchangesaccountfor85%oftotalbitcoinvolumetofiat
curencies
• Expandedsampleof30exchangesacross16countriesfrom
additionalsourcessuchasbitcoincharts.comandindividual
exchangesthemselves
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 9
Summary statistics: returns
Return Std.Dev Skewness Kurtosis ρ1 ρ2 ρ3 cross
frequency correlation
5-Minute 1.40 1.56 365.64 0.07 -0.01 0.01 0.57
Hour 1.22 -0.06 13.86 -0.07 -0.05 -0.01 0.83
Daily 1.07 0.29 3.85 -0.01 0 0.02 0.95
MakarovandSchoar,TradingandArbitrageinCryptocurrencyMarkets 10